Autonomy & WorkforceautonomyUK
End-to-End Learned Driving on Public Roads
13 June 2026
A developer demonstrated an end-to-end, learning-based driving system generalising to new cities with minimal hand-engineered rules.
Download the full case study
Branded PDF, watermarked to your account. Registration required.
Modular autonomy stacks require extensive hand-engineering and city-by-city tuning that scales poorly.
An end-to-end approach learned driving behaviour directly from data, with a foundation-model backbone intended to generalise across environments.
The system drove in cities it had not been explicitly tuned for, suggesting a path to scaling autonomy without rebuilding rules for every new market.
Vendors referenced
Sources